Ocean Modeling II. Parameterized Physics. Peter Gent. Oceanography Section NCAR

Transcription

1 Ocean Modeling II Parameterized Physics Peter Gent Oceanography Section NCAR

2 Ocean Modelling Challenges LO-RES (3 o ) O(100+ years/day) WORKHORSE (1 o ) O( years/day) HI-RES (0.1 o ) O(1 year/day) Circumference of Earth ~4x10 5 km Horizontal

3 MODEL EQUATIONS 3-D primitive equations in spherical polar coordinates with vertical z-coordinate for a thin, stratified fluid using hydrostatic & Boussinesq approximations. Continuity equation: Hydrostatic equation: Equation of state: Tracer transport : L(1) = 0 (2.8) p z = ρg (2.9) ρ = ρ(θ,s,p) ρ(θ,s,z) (2.10) t ϕ + L(ϕ) =D H(ϕ)+D V (ϕ) (2.11) D H (ϕ) =A H 2 ϕ (2.12) The common parameterization hypothesis about turbulent processes is that they mix material properties, hence the most common operator form is eddy diffusion (e.g. by spatial Laplacians) with an eddy diffusivity as the free parameter. McWilliams, 1998, Oceanic General Circulation Models, Ocean Modelling and Parameterization, NATO Science Series. D V (ϕ) = z κ ϕ, (2.13) z

4 PARAMETERIZATIONS IN CESM2 POP2 Vertical mixing (momentum and tracers) - surface boundary layer - interior Horizontal viscosity (momentum) Lateral mixing: mesoscale eddies (tracers) Overflows Submesoscale eddies (tracers) Estuary box model parameterization Solar absorption

5 VERTICAL MIXING SCHEME: K-PROFILE PARAMETERIZATION (KPP) OCEANIC VERTICAL MIXING: A REVIEW AND A MODEL WITH A NONLOCAL BOUNDARY LAYER PARAMETERIZATION W. G. Large J. C. McWilliams S. C. Doney National Center for Atmospheric Research Boulder, Colorado 1994 Copyright 1994 by the American Geophysical Union /94/94 RG Reviews of Geophysics, 32, 4 / November 1994 pages Paper number 94RG01872 Unresolved turbulent vertical mixing due to small-scale overturning motions parameterized as a vertical diffusion. Guided by study and observations of atmospheric boundary layer parameterize t X = z w X w X = K x z X where K x represents an eddy diffusivity or eddy viscosity and X = { active/passive scalars or momentum }

6 VERTICAL MIXING SCHEME: K-PROFILE PARAMETERIZATION (KPP) KPP is not just a vertical diffusion scheme because the scalars (Temp and Salinity) have non-local or countergradient terms γ x w X = K x ( z X - γ x ) KPP involves three high-level steps: 1. Determination of the boundary layer (BL) depth: d 2. Calculation of interior diffusivities: υ x 3. Evaluation of boundary layer (BL) diffusivities: K x Diffusivity throughout the boundary layer depends on the surface forcing, the boundary layer depth, and the interior diffusivity. KPP produces quite large diffusivities below the boundary layer, which mixes temp and salinity quite deep in times of very strong surface wind stress, such as strong midlatitude atmosphere storms.

7 VERTICAL MIXING SCHEME: K-PROFILE PARAMETERIZATION (KPP) 1. BL depth d is minimum depth where the bulk Richardson # (Ri b ) referenced to the surface equals a critical Richardson # (Ri cr =0.3). Ri b (d) = [ B r B(d) ]d V r V(d) 2 + V 2 t (d) B r : near-surface reference buoyancy V r : near-surface reference horizontal velocity Stabilizing buoyancy difference Destabilizing velocity shear unresolved shear V t (d) : velocity scale of (unresolved) turbulent shear at depth d Ri measures the stability of stratified shear flow. Boundary layer eddies with mean velocity V r and buoyancy B r should be able to penetrate to the boundary layer depth, d, where they first become stable relative to the local buoyancy and velocity.

8 VERTICAL MIXING SCHEME: K-PROFILE PARAMETERIZATION (KPP) 2. Calculation of interior diffusivities υ x (d) = υ x s (d) +υ x w (d) +υ x d (d) +υ x c (d) +υ x t (d) υ x : interior diffusivity at depth d (below the boundary layer) υ xs : (unresolved) shear instability υ xw : internal wave breaking υ xd : double diffusion υ xc : local static instability (convection) υ xt : tidal mixing Superposition of processes sets interior vertical diffusivity, υ x, below the surface boundary layer.

9 VERTICAL MIXING SCHEME: K-PROFILE PARAMETERIZATION (KPP) Verification example at Ocean Weather Station Papa (50 o N, 145 o W): Large et al (1994) o 1, o 1 o I 0 1 MAR AP b Mo i d,,i Figure 9. Time-depth sections of 4-day averages of observed temperatures in degrees Celsius (a) from ocean weather station (OWS) Papa during the ocean year March 15, 1961, to March 15, 1962 and (b) from the standard KPP simulation of OWS Papa.

10 JPO 2001 LAPLACIAN HORIZONTAL VISCOSITY Isotropic, spatially varying coefficient in Cartesian coordinates D(U) = (A U x ) x + (A U Y ) y D(V) = (A V x ) x + (A V Y ) y

11 ANISOTROPIC HORIZONTAL VISCOSITY t u +... = x ( A x u) + y ( B y u) t v +... = x ( B x v) + y ( A y v) Need the viscosity matrix to be symmetric in order to satisfy a number of constraints on the viscosity. Small B in u equation does not diffuse strong, but thin, equatorial currents, such as the equatorial undercurrent. Small B in v equation does not diffuse strong, but thin, western boundary currents, such as the Gulf Stream.

12 Mesoscale eddy mixing of tracers: Gent-McWilliams (GM) parameterization Lateral Mixing (Interior, Tracer Diffusion) 1990 Ocean Modelling 39 (2011) 2 9 Contents lists available at ScienceDirect 2011 Ocean Modelling journal homepage: The Gent McWilliams parameterization: 20/20 hindsight Peter R. Gent National Center for Atmospheric Research, 1 Boulder, CO, USA

13 Why is GM needed? Lateral Mixing Agulhas (Interior, Retroflection Tracer Diffusion) O(1 o ) models do not resolve the 1 st baroclinic deformation radius away from the equatorial regions, and hence lack the mesoscale turbulence which mixes temperature, salinity and passive tracers in the real ocean.

14

15 Why is GM needed? ISOPYCNAL (ρ=constant) DEPTH Isopycnal slopes are small O(10-3 ) at most Ocean Observations suggest mixing along isopycnals is ~10 7 times larger than across isopycnals. Early ocean models parameterized the stirring effects of (unresolved) mesoscale eddies by Laplacian horizontal diffusion with K H = O(10 3 m 2 /s), whereas the vertical mixing coefficient K v = O(10-4 m 2 /s). Horizontal mixing results in excessive diapycnal mixing, which degrades the ocean solution: e.g. Veronis (1975) showed that it produces spurious upwelling in western boundary current regions which short circuits the N. Atlantic MOC. Thus, was a recognized need to orient tracer diffusion in z-coordinate models along isopycnal surfaces, to be consistent with observed ocean mixing rates.

16 The GM Parameterization T u u T Ñ 2 + ( + *). Ñ = k T t r w* = -Ñ.( kñr / r ), Ñ. u* = z 0. GM (1990) proposed an eddy-induced velocity u* in addition to diffusion along isopycnal surfaces.

17 GM impacts Gent et al. (JPO, 1995): Eddy-induced velocity (v*,w*) acts to flatten isopycnals and minimize potential energy. Temperature Temperature warm Density dense light cold time Salinity Salinity salty fresh

18 Southern Hemisphere zonal wind jet Baroclinic instability produces ACC eddies that try to flatten the isopycnals and produce a MOC that opposes the mean flow MOC.

19 Southern Hemisphere zonal wind jet Mean MOC Eddy MOC Baroclinic instability produces ACC eddies that try to flatten the isopycnals and produce a MOC that opposes the mean flow MOC.

20 4 P.R. Gent / Ocean Modelling 39 (2011) 2 9 (1982) ob published most exac Impacts of GM to us. In accelerati was appli fined as t (a) Horizontal Diffusion, MOC (u) they prov actly bala was stead (b) GM, MOC (u) for ocean duced ove the mean (c) GM, MOC (u+u*) the non-a sure grad Therefore 4 o x 3 o x 20L ocean model pressure d basins, th small, eve Danabasoglu et al. (1994, Science) strong ba or the Ag zonally-av the ACC, w It is cl water from often pro was stabi The result and mode horizonta out the so high latit transport percentag shown in to just t Norwegia water for complete of my care 4 3 m

21 P.R. Gent / Ocean Modelling 39 (2011) Impacts of GM Deep Water Formation (a) Horizontal Diffusion (b) GM In (b), deep water is formed only in the Greenland/ Iceland/Norwegian Sea, the Labrador Sea, the Weddell Sea and the Ross Sea. 4 o x 3 o x 20L ocean model Danabasoglu et al. (1994, Science)

22 GM: the near-surface eddy flux GM90 is valid only in the nearly adiabatic ocean interior. Therefore, it needs to be modified near the surface as the isopycnals are nearly vertical in the mixed layer. Eddy-induced Diffusivities velocity profile diabatic Horizontal Horizontal/Isopycnal adiabatic Isopycnal FIG. 2. A conceptual model of eddy fluxes in the upper ocean. Mesoscale eddy fluxes (blue arrows) act to both move isopycnal surfaces and stir materials along them in the oceanic interior, but the fluxes become parallel to the boundary and cross density surfaces within the BL. Microscale turbulent fluxes (red arrows) mix materials across isopycnal surfaces, weakly in the interior and strongly near the boundary. The interior and the BL regions are connected through a transition layer where the mesoscale fluxes rotate toward the boundary-parallel direction and develop a diabatic component. The near-surface eddy flux scheme replaces the early approach of just tapering the GM coefficient to zero at the ocean surface. Ferrari et al. (2008, J. Climate)

23 GM summary Mimics effects of unresolved mesoscale eddies as the sum of - diffusive mixing of tracers along isopycnals (Redi 1982), - an additional advection of tracers by the eddy-induced velocity u* Scheme is adiabatic and therefore valid for the ocean interior. Acts to flatten isopycnals, thereby reducing potential energy. Eliminates any need for horizontal diffusion in z-coordinate OGCMs è eliminates Veronis effect. Implementation of GM in ocean component was a major factor enabling stable coupled climate model simulations without flux adjustments.

24 Limerick 2004 There There once was an ocean model called POP, Which occasionally used to flop, But eddy advection, and much less convection, Turned it into the cream of the crop.

25 Limerick 2000 There once was an ocean model called MOM, Which occasionally used to bomb, But eddy advection, and much less convection, Turned it into a stable NCOM.

26 GRAVITY CURRENT OVERFLOWS GREENLAND Denmark Strait ICELAND Faroe-Bank Channel SCOTLAND Latitude From Jim Price Woods Hole

27 GRAVITY CURRENT OVERFLOW PARAMETERIZATION surface Inflow : M i Interior : ρ i T i S i d s Sill Depth h e Outflow: M S W s Width Entrainment ρ E T E S E M E Shelf Break Slope, α Source : ρ S T S S S X ssb M P = M S + M E Product ρ p T p S p TRACER CONSERVATION Reduced Gravities : g s ' =(g/ρ o ) (ρ s - ρ i ) g e ' =(g/ρ o ) (ρ s - ρ e ) Based on Price & Yang (1998); described in Briegleb et al. (2010, NCAR Tech. Note) and Danabasoglu et al. (2010, JGR)

28 OVERFLOW PARAMERERIZATION MODEL SCHEMATIC Danabasoglu et al. (2010, JGR) Figure 1. A schematic of the Nordic Sea overflows. T, S, r, and M represent potential temperature, salinity, density, and volume transport, respectively. The subscripts s, i, e, and p refer to properties of the overflow source water at the sill depth, the interior water at the sill depth, the entrainment water, and the product water, respectively. The thick, short arrows indicate flow directions. The sill depth lies within the green box of raised bottom topography. The other boxes (except the orange product box) represent the regions whose T and S are used to compute the necessary densities. All parameters shown in black are constants specified for a particular overflow (Table 1). See section 2 for further details.

29 BOTTOM TOPOGRAPHY OF x1 o OCEAN COMPONENT Danabasoglu et al. (2010, JGR)

30 ATLANTIC MERIDIONAL OVERTURNING CIRCULATION (AMOC) OCN OCN* CCSM CCSM* Danabasoglu et al. (2010, JGR) in Sv

31 TEMPERATURE AND SALINITY DIFFERENCES FROM OBSERVATIONS AT 2649-m DEPTH OCN OCN* mean= 0.45 o C rms= 0.50 o C mean= o C rms= 0.13 o C o C OCN OCN* mean= 0.02 psu rms= 0.03 psu mean= psu rms= 0.03 psu psu